The dinner timer test
Britain’s universities are outpaced by China’s tech ambitions
My husband was chopping garlic when I started the timer. By the time he finished cooking, which was 45 minutes later, I had identified 29 university partnerships that raise questions about Britain’s research security. Everything I found was openly visible on the Chinese and international internet, drawn from university websites, corporate announcements, and state media reports.
In May 2018, the University of Warwick began a joint project with CloudWalk Technology, a Chinese facial-recognition company that has produced public-security applications for the Chinese state. The collaboration concerned “cross-media big-data intelligent computing” for facial recognition, re-identification, and “intelligent security.” [https://archive.ph/76ULk] In other words, technologies that allow cameras and databases to recognise the same person across cities, days, and contexts.
Two years later, in June 2020, CloudWalk was added to the U.S. Commerce Department’s Entity List for enabling high-technology surveillance of ethnic and religious minorities in China, including Uyghurs in Xinjiang. [https://archive.ph/r68Ad]
A second case: In 2018, Oxford University established a research centre inside the Suzhou Industrial Park, a strategic platform guided by state policy to advance China’s national technology and industrial goals. [https://archive.ph/zljIA] The centre, Oxford’s first in China, would work on optoelectronics, advanced materials, and AI—fields with clear dual-use applications and military relevance. [https://archive.ph/h4yxb] By 2023, the centre produced 29 patents, 70 papers, and roughly RMB 30 million (~£3–3.5 million) in research income.[https://archive.ph/6vvLg] It also established two Oxford-linked firms locally in China, one spin-off company, and had seven technologies undergoing commercialisation. [https://archive.ph/HUn03]
Functionally, the centre links Oxford research groups to the Suzhou industrial ecosystem, creating potential pathways for rapid local application. I found no public UK disclosures of export-control or dual-use assessments associated with related projects. Chinese state media, on the other hand, devoted a lengthy feature to the partnership, headlined “Fast and Furious”—celebrating the deepening relationship between Oxford and Suzhou Industrial Park.[https://archive.ph/HUn03]
Also in 2018, Arm China, the joint venture of Britain’s flagship semiconductor designer, announced its Artificial Intelligence Business Headquarters in Shanghai, including an AI-chip research institute promoting domestic self-reliance in integrated circuits. [https://archive.ph/262mV] Chinese media framed the project as advancing national chip development goals.
The name Arm China sounds like a British export, but the entity was structured from inception as a joint venture with 51% Chinese ownership, including state-linked backing. [https://group.softbank/en/news/press/20180605] Within two years, Arm’s UK parent lost operational control after the Chinese CEO refused dismissal. [https://www.reuters.com/technology/arm-china-says-its-ousted-ceo-wu-is-refusing-pack-up-2022-05-05]
These collaborations have until now received little scrutiny in English-language reporting on research security. It is unclear whether any institution named in this article conducted due diligence on their Chinese partners’ affiliations or considered the potential reputational risks. Regardless, the absence of public disclosure raises questions about transparency and accountability.
After dinner, I screened the 29 cases against NPSA Trusted Research guidance and publicly available RCAT/NSI materials (2024–2025), classifying them into four indicative risk bands: four cases (14%) critical risk, involving sanctioned or export-controlled partners; four (14%) severe risk, embedded in Chinese state industrial or diplomatic frameworks; ten (34%) substantial risk, centred on joint research, doctoral, or semiconductor projects with dual-use potential; and eleven (38%) contextual risk, mainly education or media exchanges that looked benign yet tracked with China’s strategic science-and-technology agenda.
Chronologically, nine of the twenty-nine cases (31%) occurred before the UK’s National Security and Investment Act took effect in 2022, ten (34%) took place between 2022 and 2023, and another ten (34%) emerged in 2024 and 2025. This distribution partly reflects the bias of open-source visibility: more recent collaborations are easier to trace.
This analysis is preliminary. The dataset is small, drawn entirely from public sources, and I did not record the total universe of UK–China partnerships examined during those 45 minutes.
Many of these collaborations may have been legitimate when established, but they now persist within a changed environment where AI sovereignty and technology security have become policy priorities. This is, after all, the same concern that led to Huawei’s exclusion from the UK’s 5G network in 2020, following the government’s final decision that July to remove Huawei equipment from national telecom infrastructure. [https://www.gov.uk/government/news/huawei-to-be-removed-from-uk-5g-networks-by-2027]
The strongest reason for caution is Beijing’s shift toward self-reliance in “key and core technologies” backed by a “new nationwide system” to concentrate resources. [https://cset.georgetown.edu/wp-content/uploads/t0284_14th_Five_Year_Plan_EN.pdf] China’s 14th Five-Year Plan (2021-2025) explicitly commits to “more actively plug into the global innovation network” [https://cset.georgetown.edu/wp-content/uploads/t0284_14th_Five_Year_Plan_EN.pdf], but simultaneously steers foreign collaboration toward strategic sectors—AI, semiconductors, quantum computing—where it seeks self-reliance. According to policy analysts, this represents mobilising foreign expertise for domestic strategic ends. [https://crsreports.congress.gov/product/pdf/IF/IF11684]
Since the early 2010s, military-civil fusion has been woven into virtually all major S&T initiatives—formalised through the 13th FYP Special Plan and the 2017 creation of the Central Commission for Integrated Military and Civilian Development. [https://cset.georgetown.edu/wp-content/uploads/t0163_13th_5YP_mil_civ_fusion_EN.pdf] According to policy documents, this means civilian research may serve dual purposes. [https://2017-2021.state.gov/military-civil-fusion/] This dynamic exists in Western countries too—DARPA and the UK Defence Science and Technology Laboratory also fund dual-use research. The concern is not that dual-use research exists, but the scale, centralisation, and explicit state direction of a particular country’s approach.
The challenge for the UK is that these collaborations may be channelling British expertise into strategic ecosystems whose alignment with British interests remains uncertain, even as Britain’s regulatory language has evolved. It’s possible that RCAT or the National Security and Investment Unit have reviewed some of these partnerships confidentially. The absence of public disclosure doesn’t prove absence of oversight—but it does make external accountability difficult.
Other governments have introduced tighter oversight of foreign research engagements. Rather than a universal partnership-screening regime, Australia regulates intangible technology transfers through the Defence Trade Controls Act 2012, expanded by AUKUS-related amendments in 2024. Canada’s 2023 National Security Guidelines for Research Partnerships integrate security vetting into tri-council grant assessments. Yet publicly available information does not show how effectively either framework operates in practice, or how many collaborations they have actually restricted.
This is not an argument for severing all UK-China research ties. Many collaborations deliver genuine scientific value and pose minimal security concerns. But the current approach—decentralised decision-making, minimal public disclosure, limited Chinese-language monitoring capability—leaves Britain structurally vulnerable to partnerships that may conflict with its own stated security priorities.
The question is whether the UK’s research security apparatus can scale to match the challenge. Answering that question requires the kind of systematic, Chinese-language-enabled research that goes beyond what any single researcher can accomplish over dinner.
—
A note on method: This analysis, began on 23 Oct 2025, emerged from open-source research of English and Chinese-language sources. I did not systematically document how many partnerships I reviewed and dismissed as unproblematic, which means these 29 cases represent concerning examples I found, not a comprehensive audit. The 45-minute timeframe was real, but so was the professional expertise I brought to the search—I knew where to look and what patterns might indicate risk.
These examples are illustrative, not exhaustive. They demonstrate what’s easily findable in public sources and suggest that more systematic oversight may be warranted. They don’t prove that most UK-China partnerships are problematic, but they do indicate that some deserve closer scrutiny and that current transparency mechanisms may be insufficient for external accountability.
If interested in the full 29 links found in 45 minutes, I can be contacted and may release this.
—
All entities mentioned in the piece have been given the right of reply. Only CloudWalk responded by saying:
“We believe that a discussion based on outdated and inaccurate technical descriptions cannot yield meaningful insights. Therefore, before addressing your specific questions, we must first clarify a fundamental fact:
CloudWalk Technology has long transcended the narrow and outdated label of a “facial recognition and AI surveillance company. We are a cutting-edge enterprise focused on next-generation AI Agents and industry-scale large model technology. The perceptual technologies you mentioned are merely a foundational component within the complete World Model we are building, serving for environmental interaction. Our core business is dedicated to reshaping the operational paradigms of critical industries such as finance, urban governance, and transportation through Artificial General Intelligence (AGI) technology, thereby driving global productivity transformation.”
—
We did not take a photo of the chicken because I was looking at the computer, which was a shame. Cover photo features unrelated pasta we recently ate.